How do model-independent constraints like equation-of-state bounds avoid embedding different theoretical assumptions?

# How "Model-Independent" Constraints Actually Hide Choices

You've touched on a genuine tension in cosmology: the phrase "model-independent" is misleading because *any* way of parameterizing dark energy's equation of state already encodes theoretical commitments [AG-2025.07-534].

The Parameterization Problem

When cosmologists fit data to an equation of state—say, the popular choice $w(z) = w_0 + w_a z/(1+z)$—they're claiming to be agnostic about the underlying physics. But the choice of that particular functional form *itself* is a hidden prior [AG-2024.05-113]. It privileges certain kinds of evolution over others. A different parameterization (say, polynomial in scale factor instead of redshift) would carve up the same observational space differently and yield different best-fit values.

This matters concretely: [AG-2024.05-113] shows that simple quintessence models obeying the null energy condition (a mild physical requirement) naturally prefer the same region of $w_0$–$w_a$ space as the observations do—but that preference is an *artifact of the parameterization*, not evidence for exotic physics. The true equation of state at present differs from what you'd read off the best-fit $w_0$.

Theoretical Priors as Truth-Trackers

Rather than hide these choices, better approaches make them transparent [AG-2025.07-534]. You can compare constraints derived under different parametrizations (phenomenological vs. effective field theory-inspired, for instance) and ask: which assumptions change the answer? Those are your real priors.

Even more ambitiously, you can impose *theory-informed* priors from fundamental physics—for example, swampland conjectures from string theory [AG-2026.05-232]. These constrain dark energy models and shift the inferred cosmological parameters when compared to "agnostic" flat priors. The point is to *own* the theoretical assumption rather than pretend it isn't there.

Structural Limits That Aren't About Theory

There's also a deeper issue independent of parameterization: the integral structure itself. When you observe distances (from supernovae, for instance), the equation of state is buried under a double integral over redshift [AG-2025.12-313]. This integral acts as a low-pass filter, suppressing information about rapid time-evolution. Even with perfect data, you can constrain only one dominant mode of $w(z)$—not because of hidden priors, but because of the mathematics. This is a structural limitation, not a theoretical choice.

Summary

To avoid embedding unexamined assumptions:

1. Make parameterizations explicit and compare results across choices [AG-2025.07-534].

2. State theoretical priors openly—whether they come from symmetry, stability, or string theory [AG-2026.05-232].

3. Acknowledge structural limits in what observables can constrain [AG-2025.12-313].

True model-independence would mean being honest that it doesn't exist—only more or less transparent choice-making.